I didn't test this but I'd assume it works:
A workaround you can try is making a new function that does that for you.
def says(who, what):
return renpy.say(who, speechPauses.add_speech_pauses(what))
If you are using tooling that is calling renpy.say directly, you can consider replacing renpy.say() function but renpy also uses that function, so that can lead to duplicated pauses.
If you tell me the specific situation you are dealing with, I may come up with a compromise for you that won't be as hard to work with.